Q: Dear Ethics Lawyer, I have the luck/misfortune to lead a medium-sized firm. For a while now, all the buzz has been about AI and Generative AI. The talk is seemingly split between cautionary tales about how one shouldn’t use Gen AI because of various ethical and malpractice concerns; and talk about how those firms that don’t adopt it will be failing ethical obligations to serve their clients more efficiently, and will also be left in the dust by their competitors to boot.
What are our ethical obligations? And to the extent we do invest in Gen AI, and serve our clients at less cost to them and more to us, how do we ethically charge for that? Also, if increased efficiency means billing fewer hours, have technology companies just taken over another piece of our business?
A: I think many share your weariness with the Gen AI discussion, the disruption of lawyer duties to clients and tribunals and to economic practice models, and the accompanying uncertainties. Your questions are too broad to be discussed in necessary detail in this column, but I can refer you to the latest guidance from the ABA, Formal Opinion 512, issued July 29, 2024. The opinion contains extensive discussion of evolving ethical duties under Rules 1.1 (competence), 1.6 (confidentiality), 1.4 (communication with clients), 3.1 (meritorious claims), 3.3 (candor to the tribunal), 8.4(c) (misrepresentation), 5.1 and 5.3 (supervisory and subordinate lawyers), and 1.5 (fees). Although this advice as applied to Gen AI may, like Gen AI itself, represent somewhat of a moving target, it does highlight the relevant issues.
In general, as Gen AI improves and is applied more reliably in law practice–specific ways, the duty of competence will require lawyers to consider its application to their practice to competently serve their clients, just as occurred with computer-assisted legal research services now decades ago. This will carry with it the obligation to understand the uses and benefits on the one hand, and the limitations and risks of using it, and verification of its output, on the other. This does not mean that each lawyer or firm needs to be expert in the matter, reliance on the advice of others who are is sufficient and appropriate. These duties will evolve over time with the evolution of the technology and understanding/acceptance of its application, Formal Op. 512, at 2-5; and ways that it can be used without endangering the confidentiality of client information relating to a representation. Id. at 6-7. With evolving usage of Gen AI in practice will come changing obligations to communicate about its use with both clients and tribunals who receive the product of legal work resulting from it. Id. at 8-10.
The Opinion’s discussion of fee and billing issues does foreshadow the possibility that the promise of Gen AI to create more efficient representation of clients will indeed impose more costs upon law firms with less ability to replace revenue from resulting lost billable hours. The basic premises under Rule 1.5 remain the same: lawyers may recover direct expenses on a pass-through (no mark-up) basis, but generally may not pass through “overhead” items; and lawyers’ fees must be reasonable. Using the computer-assisted legal research model as precedent, this could allow law firms (with client agreement) to pass through direct expense paid to vendors to provide Gen AI to a particular client matter (assuming that can be isolated independent of “overhead”), but not to bill for hours of lawyer time the use of Gen AI has replaced. Id. at 12-13. There are also potential implications even for “flat fee” work—the Committee suggests that if the use of Gen AI allows the lawyer to complete the same work in less time, it may be unreasonable to charge the same flat fee amount. Id. at 12. This seems somewhat counterintuitive, but as with earlier technology-based changes in practice, what is reasonable to charge in flat fees and other fee arrangements in work involving Gen AI work should evolve in the marketplace, in the same way that hourly rates have evolved and will continue to evolve to provide for profitable practice.
The Ethics Lawyer’s advice is to work through the background noise and information overload about Gen AI. If change can be thought of as a train, then it is better to ride than to stand on the tracks. Have someone in your firm who understands it at the depth suggested by Formal Opinion 512, or obtain outside professional advice about it. Be cautious in its use for now: making sure not to input client information into an iteration of Gen AI that puts that information at risk; appropriately verifying its output for client and tribunal use; and disclosing its use and obtaining client consent where needed. As legal-specific versions develop and gain acceptance, perhaps these cautions can be relaxed. Monitor the marketplace for how the direct costs of Gen AI are being recovered, and for changes in the understanding of “reasonable” rates or other billing mechanisms that will enable a profitable business model taking into account fewer billable hours and greater efficiency.